OpenAI Races to Launch AI Agent Against Rivals

OpenAI is preparing to launch its own AI agent, reportedly codenamed Aeon, at its upcoming 2026 DevDay event. This move positions OpenAI to compete directly in the rapidly expanding category of continuously running, consumer-facing AI agents. The company aims to integrate the best features of existing platforms, striving for the utility of open-source options like OpenClaw, the affordability of Meta's Muse, and the ease of use seen in agents like Instinct.

Recent months have seen intense competition in the AI agent space. Meta's Muse, released earlier this month, has quickly become a top performer, garnering 600,000 daily active users in the US and topping app store charts. Google's Gemini Spark also operates as a 24/7 consumer-facing agent, boasting over 30 external service partners. These agents are performing tasks ranging from booking travel and groceries to managing appointments and subscriptions.

Despite their utility, AI agents face significant security challenges. Past vulnerabilities have allowed attackers to gain control of accounts, and user reports include instances of agents revealing personal information like home addresses or reading private messages without authorization. OpenAI's potential offering, Aeon, will need to address these security risks effectively to succeed.

OpenAI's potential launch of Aeon is anticipated to be underpinned by its new frontier model, GPT-6 Astra, which the company has suggested marks the arrival of artificial general intelligence. This advanced AI could enhance Aeon's capabilities in areas such as cybersecurity, software engineering, and scientific research. However, OpenAI faces a disadvantage in integrating with existing user ecosystems, unlike Meta's and Google's platforms, which connect to billions of users and a wide range of services.